Tutorial 🐧

This is a gentle and lighthearted tutorial on how to use tools from SplitApplyPlot, using as example dataset a collection of measurements on penguins^[1]. See the Palmer penguins website for more information.

using CSV, DataFrames, HTTP

url = "https://cdn.jsdelivr.net/gh/allisonhorst/palmerpenguins@433439c8b013eff3d36c847bb7a27fa0d7e353d8/inst/extdata/penguins.csv"

penguins = dropmissing(CSV.read(HTTP.get(url).body, DataFrame, missingstring="NA"))
first(penguins, 6)

6 rows × 7 columns

Frequency plots

Let us start by getting a rough idea of how the data is distributed

using SplitApplyPlot, CairoMakie

specs = data(penguins) * frequency() * mapping(:species)
draw(specs)

Next, let us see whether the distribution is the same across islands.

specs * mapping(color = :island) |> draw

Ups! The bars are in the same spot and are hiding each other. We need to specify how we want to fix this. Bars can either dodge each other, or be stacked on top of each other.

specs * mapping(color = :island, dodge = :island) |> draw

This is our first finding. Adelie is the only species of penguins that can be found on all three islands. To be able to see both which species is more numerous and how different species are distributed across islands in a unique plot, we could have used stack.

specs * mapping(color = :island, stack = :island) |> draw

Correlating two variables

Now that we have understood the distribution of these three penguin species, we can start analyzing their features.

specs = data(penguins) * mapping(:bill_length_mm, :bill_depth_mm)
draw(specs)

We would actually prefer to visualize these measures in centimeters, and to have cleaner axes labels. As we want this setting to be preserved in all of our bill visualizations, let us save it in the variable specs.

specs = data(penguins) * mapping(
    :bill_length_mm => (t -> t / 10) => "bill length (cm)",
    :bill_depth_mm => (t -> t / 10) => "bill depth (cm)",
)
draw(specs)

Much better! Note the parentheses around the function t -> t / 10. They are necessary to specify that the function maps t to t / 10, and not to t / 10 => "bill length (cm)".

There does not seem to be a strong correlation between the two dimensions, which is odd. Maybe dividing the data by species will help.

specs * mapping(color = :species) |> draw

Ha! Within each species, penguins with a longer bill also have a deeper bill. We can confirm that with a linear regression

an = linear()
specs * an * mapping(color = :species) |> draw

This unfortunately no longer shows our data! We can use + to plot both things on top of each other:

specs * an * mapping(color = :species) + specs * mapping(color = :species) |> draw

Note that the above expression seems a bit redundant, as we wrote the same thing twice. We can "factor it out" as follows

specs * (an + mapping()) * mapping(color = :species) |> draw

where mapping() is a neutral multiplicative element. Of course, the above could be refactored as

ans = linear() + mapping()
specs * ans * mapping(color = :species) |> draw

We could actually take advantage of the spare mapping() and use it to pass some extra info to the scatter, while still using all the species members to compute the linear fit.

ans = linear() + mapping(marker = :sex)
specs * ans * mapping(color = :species) |> draw

This plot is getting a little bit crowded. We could instead analyze female and male penguins in separate subplots.

ans = linear() + mapping(col = :sex)
specs * ans * mapping(color = :species) |> draw

Smooth density plots

An alternative approach to understanding the joint distribution is to consider the joint probability density distribution (pdf) of the two variables.

using SplitApplyPlot: density
an = density()
specs * an * mapping(col = :species) |> draw

A Heatmap (the default visualization for a 2D density) is a bit unfortunate if we want to mark species by color. In that case, one can use visual to change the default visualization and, optionally, fine tune some arguments. In this case, a Wireframe with thin lines looks quite nice. (Note that, for the moment, we must specify explicitly that we require a 3D axis.)

using SplitApplyPlot: density
an = visual(Wireframe, linewidth=0.1) * density()
plt = specs * an * mapping(color = :species)
draw(plt, axis = (type = Axis3,))

Of course, a more traditional approach would be to use a Contour plot instead:

using SplitApplyPlot: density
an = visual(Contour) * density()
plt = specs * an * mapping(color = :species)
draw(plt)

The data and the linear fit can also be added back to the plot:

ans = visual(Contour) * density() + linear() + mapping()
plt = specs * ans * mapping(color = :species)
draw(plt)

Correlating three variables

We are now mostly up-to-speed with bill size, but we could consider how it interacts with the weight of the penguins. For that, it may be better to use a continuous color on a gradient to denote weight, and a marker shape to denote species.

ans = linear() + mapping(color = :body_mass_g) # FIXME: colorranges do not match!!
specs * ans * mapping(marker = :species) |> draw

Naturally, heavier penguins have bigger bills. We could also try and see the interplay of these three variables in a 3D plot.

#' (FIXME: implement linear in 3D)

specs3D = specs * mapping(:body_mass_g => (t -> t / 1000) => "body mass (Kg)")
plt = specs3D * mapping(color = :species) * visual(Scatter) # FIXME: add default to Makie
draw(plt, axis = (type = Axis3,))

specs3D = specs * mapping(:body_mass_g => (t -> t / 1000) => "body mass (Kg)")
plt = specs3D * mapping(color = :species, layout = :sex) * visual(Scatter)
draw(plt, axis = (type = Axis3,), figure = (resolution = (1200, 400),))

This page was generated using Literate.jl.